Up to four models
MiniMax-M3 vs GPT-5.4 Mini
On the index (GPT-4 = 100), MiniMax-M3 scores 147 and GPT-5.4 Mini 147. Side by side on the same third-party tests, with price and context. The link in your address bar shares this exact comparison.
147131–164
WeightsOpen
ReleasedJun 2026
Price in / out$0.30 / $1.2
Context1.05M
Core tests taken4 of 9
GPT-5.4 Mini
147131–163
WeightsClosed
ReleasedMar 2026
Price in / out$0.75 / $4.5
Context400K
Core tests taken5 of 9
MiniMax-M3 and GPT-5.4 Mini are too close to call (0 points apart, within the uncertainty), ahead on 2 of 3 shared core tests, and costs 3.2× less.
Test by testthird-party results
Every result as a table
| Test | MiniMax-M3 | GPT-5.4 Mini |
|---|---|---|
| Reasoning & knowledge | ||
| GPQA Diamond (Epoch AI) | 90.9% | 86.9% |
| GPQA Diamond (Vals AI) | 92.7% | 83.1% |
| MMLU Pro | 84.2% | 84.5% |
| MMMU Pro | 81.2% | 79.3% |
| Maths | ||
| AIME-style maths (OTIS mock) | 71.1% | 88.9% |
| Chess puzzles | 17.0% | 23.0% |
| Coding | ||
| Terminal-Bench 4.0 | 1.0% | 2.5% |
| Code Migration | 19.9% | 12.9% |
| LiveCodeBench | 82.2% | 81.5% |
| SWE-bench | 75.0% | 73.0% |
| Vibe Code Bench v1.1 | 47.6% | 48.0% |
| Agents | ||
| DTBench | 78.9% | 80.0% |
| LMCA | 33.7% | 40.8% |
| Mystery game puzzles | 8.0% | 11.0% |
| Professional work | ||
| Finance Agent (v2) | 48.3% | 45.4% |
| CorpFin v2 | 68.1% | 60.9% |
| EMB | 47.8% | 45.4% |
| Legal Research Bench | 29.8% | 12.5% |
| MortgageTax | 68.4% | 63.5% |
| Tax Agent Bench | 20.6% | 8.8% |
| TaxEval v2 | 72.7% | 71.2% |
Tests at least two of these models have taken. Bold rows are the core tests.